With the rapidly growing database of gene sequence information, there is a need for techniques that can rapidly and efficiently attribute function in vivo. To date, in a mammalian model, the only solution has been the generation of transgenic knockout animals. Knockout animal models such as mice and rats are a powerful tool in studying the role genes play during development. In these models, a targeted gene is made non-functional, resulting in an altered phenotype. This phenotype may be indicative of the function of the gene and the role it plays during development. However, the transgenic knockout method suffers from several significant drawbacks, including technical difficulty, time required, the limitation to single gene targets, and the inability to uncover time-dependent secondary phenotypes.
Currently, the most widely used approach in which such knockouts are created is disruption of the gene by genetic recombination with exogenous DNA (Mansour (1990) Genet. Anal. Tech. App. 7:219-227; Robertson (1991) Biol. Reprod. 44:238-245; Zimmer (1992) Ann. Rev. Neurosci. 15:115-137). In this process oocytes collected from an animal stimulated to superovulate are transfected with the exogenous DNA. Recombination occurs, resulting in an oocyte with a non-functional gene. The oocyte is then fertilized, and implanted back into the animal. All subsequent progeny of these transformed animals will contain a disrupted target gene. However, this procedure suffers from several significant drawbacks. The transgenic knockout method is expensive, complex, and time consuming. In addition, the recombination event is limited to a point early in development. Finally, the technique is not capable of targeting more than one gene at a time, except as mating events between the eventual progeny of two separate recombination knockout animals.
Another method which has been used to create animal knockouts is the treatment of embryos in vitro with antisense RNA (Brice et al. (1993) Dev. Genet 14:174-184) or DNA (Brice et al. (1993) Dev. Genet. 14:174-184; Ochiya et al. (1995) J CellBiol. 130:997-1003; Chen et al. (1995) Biol. Reprod. 53:1229-1238; Augustine et al. (1995) Teratol. 51:300-310; Stutz et al. (1997) Mol. Cell. Biol. 17:1759-1767). Antisense approaches have the advantage of disrupting gene expression at the level of mRNA, resulting in the equivalent of a homozygous knockout phenotype. However, the methodology currently used is also expensive, complex, and time-consuming, in that it requires manipulation and growth of embryos in an artificial in vitro environment. It would be a distinct advantage to be able to administer oligonucleotides to pregnant mice in order to generate phenotypic knockouts directly, rather than treat embryos in vitro. The obvious barrier to this is the placenta.
Thus, improved methods of producing knockout models and of defining and studying gene function are still needed.
Antisense oligonucleotides have shown promise as candidates for therapeutic applications for diseases and disorders resulting from expression of cellular genes (see, e.g., WO 95/09236, WO 94/26887, and PCT/US/13685). Synthetic antisense oligonucleotides have also been demonstrated to be useful tools in inhibiting a wide variety of viruses (see, e.g., Agrawal (1992) Trends Biotech. 10: 152-158). The development of various antisense oligonucleotides as therapeutic and diagnostic agents has recently been reviewed by Agrawal and Iyer (Current Opinion Biotech. (1995) 6:12-19). More recently, the antisense approach has been found to be useful for therapeutic treatment in vivo (see e.g., Agrawal (1996) TIBTECH 14:376-387; Monia et al. (1996) Nature Medicine 2:668-675; Crooke et al. (1996) Ann. Rev. Pharmacol. Med. 36:107-129).
However, neither the treatment of embryos for viral infections or aberrant gene expression nor the modulation of embryonic genes or foreign genes harbored by embryos have heretofore been successfully accomplished in utero via the antisense approach. Such demonstrations become important when devising noninvasive therapeutic methods and informative in vivo models for human disease. It has been shown that phosphorothioate-modified oligonucleotides injected into pregnant female mice are not toxic or teratogenic to embryos in a non-specific way (Gaudette et al. (1993) Antisense Res. Dev. 3:391-397).
Accordingly, there remains a need for effective gene-specific antisense oligonucleotide therapy suitable for treatment in mammalian embryos.